基于BP神经网络技术的路段实际通行能力
The Road Section Actual Capacity Based on the BP Neural Network
-
摘要: 交通工程领域对于道路通行能力的计算的研究主要是在常态条件下对理论通行能力修正系数的研究,但路段上的随机因素对于实际通行能力的影响也是不可忽视的.为此,对现行的路段实际通行能力计算方法进行了分析,并确定了影响实际通行能力的各种因素,对其随机性进行分析.在此基础上,结合BP神经网络模型能够很好地体现出随机性的影响,对通行能力的影响因素和实际通行能力之间的函数关系进行逼近,得到相应的拟合公式.依据已有的样本,建立了实际通行能力计算的神经网络模型,结果采用神经网络模型计算的实际路段的通行能力与实测数据误差最大为4.09%,因此神经网络模型适宜于路段实际通行能力的计算.Abstract: The calculation study on the road capacity mainly focuses on the study on the correcting coefficients of the theoretical road capacity under the normal condition, but the effect of the stochastic factors on the road section on the practical capacity cannot be rechoned with, either. Therefore, the authors analyze the calculation method of the actual capacity of the road section, decide the various factors affecting the actual capacity and analyze their random. On this basis, by means of the flexibility in use of BP neural network, the easy understanding in method, and the good embodiment of the effect of random, the functional relationship between the influencing factors and the actual capacity is approached in the model so that the equation of fitting formula is obtained. According to the existing sample, the neural network model for the calculation of actual capacity is established and tested.